Rapidly Changing Data

Understanding the origin of the company were data is written once and read many times will this technology be successful in a high volume rapidly changing data base environments.

How would this work with a managed database system such as DB2, Oracle, SQL, ProgressSQL, Access or any other database.

John,

Most of our dsNet testing and optimization so far has been with unstructured data, especially large unstructured data -- like content -- so, we have a much better sense of performance with unstructured data vs. databases. 

Because the dsNet supports a standard iSCSI/block interface, these database applications will run.  You do have to consider the performance, though.  At a minimum, a dsNet works well for storing log files, archives and snapshots.

If your database performance requires low latency and the database needs to do a lot of small reads and writes, you probably won't want to use the dsNet for primary storage without setting up some kind of cache (on a local hard drive, array or in RAM). 

The bottom line is that your performance mileage may vary, so you'll want to test your application first.  We've been working with Medusa Labs who have been running their 'Pain' and 'Maim' test suites on a dsNet.  As evidenced by their names, these test suites are quite rigorous.  So databases applications generally will run; you just need to check the performance -- especially if you are considering a geographically dispersed dsNet (as opposed to a local dsNet which would perform better with a database application).

Let us know if you end up doing any testing with database applications as it will be interesting to hear about your results.

Regards,

Chris Gladwin